Cutaneous Lupus Erythematosus and Cancer Risk
Cutaneous Lupus Erythematosus and Cancer Risk
We conducted a study to examine overall and specific cancer risks in a population-based cohort of patients diagnosed with CLE in Sweden. We included in total 3663 individuals from the National Patient Register (NPR) with a diagnostic code of CLE during the period 1 January 1997–31 December 2007. CLE was defined according to the relevant code in the 10th revision of the International Classification of Diseases (ICD-10): L93 [L930 DLE, L931 SCLE and L932 other local lupus erythematosus (LE)].
From Statistics Sweden a matched control cohort from the general population was generated (n = 10 989), i.e. three control subjects to each CLE patient case (71 of the collected control subjects died during the matching process before actually being used as reference because of the sampling procedure). The control subjects did not have an earlier diagnosis of CLE; otherwise their health status was representative of that of the general population. They were matched for age, sex and county of residence.
Since 1964 the Swedish National Board of Health and Welfare has collected data on individual discharges in the NPR, and from 1987 the register has covered 100% of public, inpatient care in Sweden. Since 2001, specialists (general practitioners not included) working in outpatient clinics have also reported to the register, with coverage of about 80%. Validations have shown a correct ICD code at the four-digit level in 86% of all discharge diagnoses in the NPR, and more than 98% of discharge diagnoses were coded technically correctly in 2007 according to the Swedish National Board of Health and Welfare's own study. No validation of the register has been performed for the specific diagnosis of CLE.
The register contains basic demographic factors (PRN, sex and place of residence), hospital identification and medical data (one main and up to five secondary discharge diagnoses). These diagnoses have been coded according to ICD-10 from 1997.
Information on cancer outcomes was obtained through linkage of the patient discharge data with the National Cancer Register data for the period 1958–2007. The SCR, founded in 1958 by the Swedish National Board of Health and Welfare, covers the entire Swedish population (about 9·5 million). During the follow-up of this study, the 7th revision of the International Classification of Diseases (ICD-7) was used to classify all incident cancers. Nonmelanoma skin cancer (NMSC) included SCCs, but not basal cell carcinomas. Chronic lymphocytic leukaemia (ICD-7: 204.1) was included among the lymphomas.
Most cases of newly diagnosed cancers are registered by more than one source (clinicians, pathologists, cytologists) and reporting is compulsory. The completeness of the register has been reported to be over 96%. Data include information on tumour characteristics (site of tumour, histological type, basis and date of diagnosis) and follow-up data (date and cause of death and date of migration). Information on previously nondiagnosed cancers from death certificates is not included in the register.
To be able to calculate person-years at risk the patient data set was also linked to the Swedish Cause of Death Register, a register containing cause of death and the date for all deceased Swedish residents since 1961. The person-years at risk were calculated from the date of the first CLE diagnosis for each individual until the date of the first cancer diagnosis for each specific cancer, date of death or the end of the observation period (31 December 2007), whichever occurred first.
The Regional Ethical Review Board in Stockholm, Sweden approved this study.
Hazard ratios (HRs) were calculated using Cox proportional hazards regression models, adjusted for follow-up time, comorbidity with SLE, sex, CLE subset and different age categories (< 50 years, 50–64 years, > 65 years). HRs (representing the relative risk of being diagnosed with cancer in the cohort of patients with CLE compared with the control cohort) and 95% confidence intervals (CIs) were calculated for cancer overall and for all major cancer types. Patients diagnosed with the same type of cancer before CLE diagnosis were excluded in the cancer subgroup analysis.
Logistic regression was used to examine the prevalence of cancer prior to CLE diagnosis, generating relative risks as prevalence odds ratios (PORs) with 95% CIs.
The χ test and Student's t-test were used when comparing means between groups.
The data were analysed using the statistical package SAS® software, version 9·2 (SAS Institute Inc., Cary, NC, U.S.A.). Differences were considered significant at P < 0·05.
Materials and Methods
Study Population
We conducted a study to examine overall and specific cancer risks in a population-based cohort of patients diagnosed with CLE in Sweden. We included in total 3663 individuals from the National Patient Register (NPR) with a diagnostic code of CLE during the period 1 January 1997–31 December 2007. CLE was defined according to the relevant code in the 10th revision of the International Classification of Diseases (ICD-10): L93 [L930 DLE, L931 SCLE and L932 other local lupus erythematosus (LE)].
From Statistics Sweden a matched control cohort from the general population was generated (n = 10 989), i.e. three control subjects to each CLE patient case (71 of the collected control subjects died during the matching process before actually being used as reference because of the sampling procedure). The control subjects did not have an earlier diagnosis of CLE; otherwise their health status was representative of that of the general population. They were matched for age, sex and county of residence.
The Swedish National Patient Register
Since 1964 the Swedish National Board of Health and Welfare has collected data on individual discharges in the NPR, and from 1987 the register has covered 100% of public, inpatient care in Sweden. Since 2001, specialists (general practitioners not included) working in outpatient clinics have also reported to the register, with coverage of about 80%. Validations have shown a correct ICD code at the four-digit level in 86% of all discharge diagnoses in the NPR, and more than 98% of discharge diagnoses were coded technically correctly in 2007 according to the Swedish National Board of Health and Welfare's own study. No validation of the register has been performed for the specific diagnosis of CLE.
The register contains basic demographic factors (PRN, sex and place of residence), hospital identification and medical data (one main and up to five secondary discharge diagnoses). These diagnoses have been coded according to ICD-10 from 1997.
The Swedish Cancer Register
Information on cancer outcomes was obtained through linkage of the patient discharge data with the National Cancer Register data for the period 1958–2007. The SCR, founded in 1958 by the Swedish National Board of Health and Welfare, covers the entire Swedish population (about 9·5 million). During the follow-up of this study, the 7th revision of the International Classification of Diseases (ICD-7) was used to classify all incident cancers. Nonmelanoma skin cancer (NMSC) included SCCs, but not basal cell carcinomas. Chronic lymphocytic leukaemia (ICD-7: 204.1) was included among the lymphomas.
Most cases of newly diagnosed cancers are registered by more than one source (clinicians, pathologists, cytologists) and reporting is compulsory. The completeness of the register has been reported to be over 96%. Data include information on tumour characteristics (site of tumour, histological type, basis and date of diagnosis) and follow-up data (date and cause of death and date of migration). Information on previously nondiagnosed cancers from death certificates is not included in the register.
To be able to calculate person-years at risk the patient data set was also linked to the Swedish Cause of Death Register, a register containing cause of death and the date for all deceased Swedish residents since 1961. The person-years at risk were calculated from the date of the first CLE diagnosis for each individual until the date of the first cancer diagnosis for each specific cancer, date of death or the end of the observation period (31 December 2007), whichever occurred first.
Ethical Committee
The Regional Ethical Review Board in Stockholm, Sweden approved this study.
Statistical Analysis
Hazard ratios (HRs) were calculated using Cox proportional hazards regression models, adjusted for follow-up time, comorbidity with SLE, sex, CLE subset and different age categories (< 50 years, 50–64 years, > 65 years). HRs (representing the relative risk of being diagnosed with cancer in the cohort of patients with CLE compared with the control cohort) and 95% confidence intervals (CIs) were calculated for cancer overall and for all major cancer types. Patients diagnosed with the same type of cancer before CLE diagnosis were excluded in the cancer subgroup analysis.
Logistic regression was used to examine the prevalence of cancer prior to CLE diagnosis, generating relative risks as prevalence odds ratios (PORs) with 95% CIs.
The χ test and Student's t-test were used when comparing means between groups.
The data were analysed using the statistical package SAS® software, version 9·2 (SAS Institute Inc., Cary, NC, U.S.A.). Differences were considered significant at P < 0·05.
